Patients with tetraplegia, or paralysis of all four limbs, are severely disabled. Surveys have found that restoration of upper-limb function is a high-priority for such patients. Brain-Machine Interfaces (BMIs) can eventually restore upper-limb reaching and grasping function by seamlessly merging the computational power of the brain with artificial prosthetic systems. A major challenge is robust translation of BMI technology to patient care. Two well-recognized limitations of current approaches are instability of recordings and the lack of proprioceptive feedback signals. This research proposal aims to conduct a pilot clinical study to test electrocorticography (ECoG) based control of an anthropomorphic exoskeleton in tetraplegic patients with residual proprioception. We specifically seek to translate BMI technology to the significant subset of tetraplegic patients with intact sensation (e.g. amyotrophic lateral sclerosis or incomplete spinal cord injury). Our approach would capitalize on both the well-recognized stability of ECoG recordings and the natural sensory feedback generated by passive movements of the subject's arm by the exoskeleton. The proposed research should greatly advance translational efforts by optimizing control under conditions that maximize neural learning mechanisms and provide natural sensory feedback.